EViews.com

Hi,

I have an x and y variable who seem to be strongly linked until my x variable goes above a certain number, where the relationship breaks down (a breakpoint), and a new relationship begins (curve is much less steep). Is it possible to find the optimal breakpoint using EViews? And if so how would I go about doing it.

Thanks!

There are many breakpoint tests, and some breakpoint estimation techniques in EViews 8. Probably the best place to start is to look up "breakpoint" in the User Guide index.

When using the Quant-Andrews test when it asks what variables do you wish to vary across the breakpoint - what is the effect of using your x variable and a constant vs just your x variable? It seems to change the result substantially, but I cant figure out which one is correct

The latter constrains the intercept to be the same across regimes.